HAL Id: halshs-00264533https://halshs.archives-ouvertes.fr/halshs-00264533

Submitted on 21 May 2008

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Combining Optical and SAR Data to MonitorTemprerate Glaciers

Emmanuel Trouvé, Gabriel Vasile, Michel Gay, Pierre Grussenmeyer,Jean-Marie Nicolas, Tania Landes, Mathieu Koehl, Jocelyn Chanussot,

Andreea Julea

To cite this version:Emmanuel Trouvé, Gabriel Vasile, Michel Gay, Pierre Grussenmeyer, Jean-Marie Nicolas, et al.. Com-bining Optical and SAR Data to Monitor Temprerate Glaciers. IEEE Geoscience And Remote SensingSymposium, Jul 2005, Seoul, South Korea. pp.2637-2640. �halshs-00264533�

Combining Optical and SAR data to monitor temperate glaciers

E. Trouve1, G. Vasile1, M. Gay2, P. Grussenmeyer3, J.-M. Nicolas4

T. Landes3, M. Koehl3, J. Chanussot2 and A. Julea1

1 : Laboratoire d’Informatique, Systeme, Traitement de l’Information et de la Connaissance

Universite de Savoie - ESIA - BP 806 - F-74016 Annecy Cedex - FRANCE

Tel: 33+ 450 096 548 - Fax: 33+ 450 096 559 - {emmanuel.trouve|gabriel.vasile}@univ-savoie.fr

2 : Laboratoire des Images et des Signaux - INP Grenoble - BP 46 - 38402 Saint-Martin-d’Heres - FRANCE

Tel: 33+ 476 826 256 - Fax: 33+ 476 826 384 - Michel.Gay@lis.inpg.fr

3 : Equipe Photogrammetrie et Geomatique, MAP-PAGE UMR 694 - INSA de Strasbourg

24, Bd de la Victoire, 67084 Strasbourg Cedex - FRANCE

Tel: 33+ 388 144 733 - pierre.grussenmeyer@insa-strasbourg.fr

4 : Departement TSI - Ecole Nationale Superieure des Telecommunications - 46, Rue Barrault

75013 Paris - FRANCE - Tel: 33+ 145 818 129 - Fax: 33+ 145 813 794 - nicolas@tsi.enst.fr

Abstract

Monitoring temperate glaciers activity becomes more andmore necessary for economical and security reasons andas an indicator of the local effects of global changes. Thispaper presents the beginning of a three year project whichregroups four laboratories to develop and combine specificmethods to extract information from optical and radar re-mote sensing data. Preliminary results are presented onthree different information sources: airborne photography,space-borne multi-spectral images and SAR interferome-try, which respectively allow the compution of high reso-lution DTM, the detection of glacial lakes and the mea-surement of glacier surface velocity. Results obtained ontwo glaciers located in the French Alps are compared andvalidated with ground measurments.

Keywords: Glacier monitoring, Photogrammetry, Digi-tal Terrain Models, SAR interferometry

1 Introduction

To monitor glacier evolution by remote sensing requiresthe development of specific image processing methods toextract information from the huge amount of data ac-quired by airborne and space-borne systems. Optical andSynthetic Aperture Radar (SAR) images are complemen-tary information sources which can be combined to derivedifferent measurements such as the height to build Digi-tal Terrain Models (DTM), or the displacement betweentwo acquisitions to obtain glacier surface velocity fields.Large data sets are now available over several glaciers inthe Alps: 20 years of airborne photography, more thanten years of multi-spectral and SAR images acquired re-

spectively by SPOT and European Remote Sensing (ERS)satellites. But such data are rarely used because of theprocessing difficulty when methods are not adapted to theglacier context: high relief topography, large displacementwithin a few days. . .

This paper presents the first results of a 3 year projectwhich started in 2004 and regroups 4 laboratories special-ized in optical and SAR image processing and fusion tech-niques. A global strategy illustrated in Fig. 1 is proposedto fill the gap between the increasing number of imagesavailable and the information useful to model glacier evo-lution and to measure the risk in the surrounding areas.The three main research axes are:

• the construction of DTM and ortho-images from highresolution optical images, and the computation of dif-ferences after one or several years to detect changessuch as volume variations, glacier retreat, lakes ap-pering/disappearing...

• the computation of differential SAR interferogramsby subtracting the topography provided by the pre-vious results to obtain displacement fields over a fewdays only.

• the fusion of the measurements provided by the twofirst axes, and features detected in optical or SARdata, allowing their tracking and the computation ofrisk maps.

Two glaciers, “Mer de Glace” and “Argentiere” locatedin the French Alps near Mont-Blanc have been selected astest sites to gather optical and SAR data. Ground mea-surements, which are regularly acquired in this area, pro-vide complementary information useful either to increasethe precision of remote sensed measures or to compare andvalidate experimental results.

0-7803-9051-2/05/$20.00 (C) 2005 IEEE

Optical

Images

DEMOrtho-

images

Mass

balance

Annual dis-

placement

Axis 1

Photogrammetry

Rectification

DEM difference

Disparity measures

Raw SAR

data

SLC images

Amplitude Phase Coherence

Velocity

field

SAR processing

Dif. Interf. generation

Filtering

Phase unwrapping

Axis 2

Target dis-

placementChanges

Optical SAR fusion

Target detection

Axis 3

Figure 1: Flowchart of the processing steps to extract informa-tion from optical and SAR data to monitor glacier evolution.

2 Optical Data

Two different sources of optical data are used with a dif-ferent purpose : airborne photography which allows thecomputation of very high resolution DTM of the glacierssurface and their borders, and space-borne multi-spectralimages with a lower spacial resolution but a strong classi-fication potential.

2.1 Airborne photography

For several decades, airborne photography has been animportant source of information over a large number ofglaciers. In the French Alps, the main glaciers of threedifferent regions have been covered by airborne photogra-phy every 3 years, between 1975 and 1995 (Tab. 1).

1975 1976 1978 1979 1981 1982 1983 1984Mont-Blanc × × ×

Savoie × × × ×

Oisans × × × ×

1985 1986 1988 1989 1990 1991 1992 1995Mont-Blanc × × ×

Savoie × × ×

Oisans × ×

Table 1: Glacier airborne photography of 3 different regionsin the French Alps.

In the framework of the project described by this paper,a few series of photos have been selected over the Mer deGlace and the Argentiere glaciers. Along these glaciers, aset of ground control points (GCP) measured by geodeticGPS with 10 cm accuracy are available. The 9 × 9 inchnegatives scanned at a resolution of 15 microns yield seriesof 15360× 15360 pixels images which cover a glacier withabout 80% overlap. On the 1995 acquisition over the Merde Glace glacier (Fig. 2), the pixel size in the originalphotos is 36 cm on the bottom of the glacier (1000m ASL)and 18cm on its top (2800m ASL). The flying height was4650m ASL.

In the first step of the processing, the digital images areorientated by using the bundle block aerial triangulation

(AT) technique [1]. On the Mer de Glace, approximatively3700 AT points were computed, including 2000 three raypoints (the same point measured in three following im-ages), 500 four ray points and about 200 five ray pointsthanks to the important overlap. The RMS on the com-puted AT points is 20 cm in planimetry and 30 cm inaltimetry. The high redundancy justifies the accuracy.

Then the images are used to compute the DTM by Pho-togrammetric Matching Techniques (KLT software pack-age) based on correlation of image patches in two or moreimages. Break-lines and structure lines are defined bystereoscopic manual measurements. The resolution variesbetween 2 to 5 meters depending on the slopes. Theglobal result is obtained with 80% automatic measurement(semi-automatic process guided by the operator who den-sifies weak areas by stereoscopic measurements). About20% are fully manually measured when no results are ob-tained from the matcher in areas with important slopesor when images present poor contrast. Specific difficultiesalso arise when matching the crevasses: manual pointshave to be taken at the top and at the bottom. The re-sult is controlled by stereo-viewing (superimposition of theDTM on the selected stereo-pairs). Fig. 3 illustrates theresult obtained with the 1995 photos of the Mer de Glace.

Figure 2: Layout of the set of the Mer de Glace photos from1995 processed by photogrammetric methods.

2.2 Space-borne multi-spectral images

When the weather conditions makes their acquisition pos-sible (usually during the summer season), multi-spectralsatellite images are very useful to detect interesting fea-tures such as lakes, glacier borders... With 10 meter res-olution for panchromatic images and 20 meters for multi-spectral images, the SPOT data used in this project (Tab.2) cover large areas and allow classification of differentland cover by their spectral signatures.

An original approach is proposed to automatically de-tect and characterize mountain lakes in multi-spectralSPOT images [2]. The classification is based on the Spec-tral Angle Mapper (SAM) algorithm which is robust tothe illumination variations since angles between vectorsare independent from their length. To reduce false alarms

Figure 3: Mer de Glace glacier 3D-model (photos draped onthe DTM) computed from the 1995 photos.

Satellite Date Band(s) Latitude Longitude

SPOT 1 07/20/00 MS 45◦55′34′′N 6◦53′15′′ESPOT 1 07/31/00 MS 45◦55′29′′N 6◦53′18′′ESPOT 1 08/26/00 PAN 46◦04′53′′N 6◦52′42′′ESPOT 2 08/29/00 PAN 46◦11′53′′N 6◦50′32′′ESPOT 2 09/25/00 MS 45◦55′26′′N 6◦52′12′′ESPOT 4 10/04/00 MS 45◦55′26′′N 6◦51′50′′E

Table 2: Panchromatic (PAN) and Multi-Spectral (MS) im-ages acquired by SPOT satellites over the Mont-Blanc area

and miss-detections, the classification results are aggre-gated with spatial information resulting from a segmenta-tion based on irregular pyramid techniques. A comparisonof the results with ground surveys show that the SAManalysis of SPOT data is an efficient tool to detect andmonitor glacial lakes and to provide information to a riskprevention management system.

3 SAR Data

Satellite SAR images are used more and more to observeglaciers in particular because of two great advantages: theactive SAR sensor acquires images independently from theweather conditions and it measures both amplitude andphase of the backscattered signal. At the resolution of thedata which were up to now available (about 20 m withERS images), the amplitude is often difficult to use toextract precise information on specific features, but thenext generation of SAR satellites should provide meterresolution images and fully polarimetric data, which willbe very useful to detect different features and backscat-tering mechanisms. The phase includes a geometric de-terministic component which makes SAR interferometryfeasible and offer the opportunity to measure the glacierdisplacement between two acquisitions on repeat passes.Several approaches can be used to separate the topograph-ical fringes from the displacement phase signal [3].

In this project, 10 raw SAR images from ERS 1 andERS 2 have been selected to study the feasibility of SARinterferometry in order to extract surface velocity fields ofthe studied glaciers. This data set presented in Tab. 3 in-cludes different time intervals (1 day with tandem couples,3 and 35 days), different seasons (spring and summer) andascending and descending passes.

Table 3: ERS SAR raw data acquired over Mont-Blanc area.

The processing techniques used to form interferogramsconsist of computing single look complex images (SLC)either by the standard DIAPASON softwares [4] or bya new technique based on beam-forming in the temporaldomain followed by interferometric registration performedby RAT software [5]. The SAR processing in the tempo-ral domain has the advantage to be more flexible in theadjustment of several parameters, including taking the lo-cal height into account which will become necessary whenhigher resolution images will become available.

The following steps of the processing consist of:

• reducing the phase noise and obtaining a robust co-herence estimation by using amplitude driven adap-tive neighborhoods and local phase slope compensa-tion to ensure the signal stationarity over the estima-tion windows [6],

• unwrapping the phase by a weighted least square al-gorithm [7] using the coherence estimate as weights,

• transforming the projection of the displacement pro-vided by the unwrapped interferogram into a velocityfield, with conventional assumptions: uniform speed,parallel to the surface gradient.

The different steps are illustrated in Fig. 4 with the 1996tandem couple. The experimental results obtained onthese descending images show a high preservation of thecoherence within one day intervals in the spring season.Comparisons have been performed with ground measure-ments providing one year displacement on several points.They show a good agreement between the InSAR speedmeasurements and the average speed on the Mer de Glaceand the Argentiere glaciers [8].

By using DEMs with different resolutions, several simu-lations have been performed, in order to asses the glaciervisibility in ERS ascending and descending acquisitions.

a) b)

c) d)

e) f)

Figure 4: March 10-11 1996 Interferogram over Argentiereglacier: a) coherence, b) phase e) amplitude after initial multi-looking; c) coherence d) phase filtered by adaptive neighbor-hood technique, f) unwrapped phase.

Despite the good orientation of the glaciers with respectto ERS orbits, only 10 % of the Mer de Glace are visiblein the ascending passes (due to foldover of strong relief).

On the descending ERS-1 pair acquired in the summerof 1991 at a 3 day interval, the standard interferomet-ric corregistration by patches algorithm has been applied.The obtained interferograms have a good overall quality,showing a high level of coherence on the nearby moun-tains and in the Chamonix valley. However the coherenceis very low on the studied glaciers. The loss of interfero-metric coherence can be explained either by an importantchange of the glacier surface state and altitude, or by astrong glacier displacement that would affect the globalcorregistration algorithm. In order to investigate whichassumption holds, we introduced a controlled displace-ment on a single ERS SLC image and built the inter-

ferogram between the original and the displaced SLC. Byvarying the induced displacement, we studied the robust-ness of the corregistration algorithm. The results presenta correct coherence level even for much larger displace-ments than the one of the studied glaciers within 3 dayintervals. In conclusion, with 3 days intervals during thesummer season, the variations of the glacier surface stateand altitude, in temperate alpine glaciers such as those ofthe Chamonix valley, are stronger than the standard SARintereferometric chain tolerances for ERS data (C band).

4 Conclusions and perspectivesThe preliminary results presented in this paper on twoglaciers in the Mont-Blanc area show the benefits of usingboth optical and SAR remote sensing data to regularlyobtain measurements on the whole glacier surface. Thehigher resolution of recently lunched and future opticaland SAR satellites (SPOT-5, RADARSAT-2, CSK...) willincrease the potential of remote sensed data to monitorglacier evolution. The next steps in the project describedin this paper will focus on the fusion of the obtained mea-surements and extracted features to derive higher levelinformation such as hazard factors and risk maps.

References

[1] J.-B. Henry, J.-P. Malet, O. Maquaire, and P. Grussenmeyer.The use of small-format and low-altitude aerial photos for therealization of high-resolution DEMs in mountainous areas: appli-cation to the Super-Sauze earthflow (Alpes-de-Haute-Provence,France). Earth Surface Processes and Landforms ISSN 0197-9337, 27(12):1339–1350, 2002.

[2] J. Chanussot, M. Gay, and P. Bertolino. Validated spectral anglemapper algorithm for glacial lake detection: comparative studybetween spot and terrestrial measurements. In IGARSS 05,Seoul, Korea, 2005.

[3] A.I. Sharov, K.Gutjahr, F. Meyer, and M. Schardt. Methodicalalternatives to the glacier motion measurement from differen-tial SAR interferometry. In Photogrammetric Computer Vision,ISPRS Technical Commission III Symposium 2002, Graz, Aus-tria, volume XXXIV, pages A–324 ff, 2002.

[4] D. Massonnet and F. Adragna. Description of the DIAPASONsoftware developped by CNES, current and future applications.In FRINGE’96 Workshop, Zurich, Switzerland, page CDROM,1996.

[5] Andreas Reigber, Stephane Guillaso, Olaf Hellwich, Marc Jager,and Maxim Neumann. Polinsar data processing with RAT(Radar Tools). In Proc. of PolInSAR’05, Frascati, Italy, pageCDROM, 2005.

[6] G. Vasile, E. Trouve, M. Ciuc, and V. Buzuloiu. General adap-tive neighborhood technique for improving SAR interferometriccoherence estimation. Journal of Optical Society of America(JOSA-A), 21(8):1455–1464, 2004.

[7] D. C. Ghiglia and L. A. Romero. Robust two-dimensionalweighted and unweighted phase unwrapping that uses fast trans-forms and iterative methods. J. Opt. Soc. Am. A, 11(1):107–117,1994.

[8] L. Bousquet, M. Gay, B. Legresy, G. Vasile, and E. Trouve. Ve-locities field of mountain glacier obtained by synthetic apertureradar interferometry. In IGARSS 04, volume II, pages 1132–1135, Anchorage USA, 2004.